Dear SPMers,
we have some problems to analyse fMRI Data of a single 80-year old aphasic
lady. In short: she was investigated twice with a simple overt naming
paradigm before (T1) and after (T2) intensive language therapy. Naming
comprised 80 pictures that alternated with rest-trials (blocks of 5 for
each condition). TR was 6 seconds during a sparse sampling procedure(3 sec
seeing & naming, approx. 2 secs aquisition of a single volume).
Furthermore we investigated 3 healthy 80-year old women twice with the
same paradigm (actually it took us some months to get this data, there are
not many of them around in our area...).
Since there is not much literature on how to deal with patient pre-post
data in fMRI here are some questions for the experts:
1. Would it be appropriate to stack pre- & post date in a single modell,
since there might be differences in the general activation at both times?
2. Leger et al (2002, Neuroimage) calculated compound contrasts to compare
pre-post data of a single patient as follows: (Name2-Rest2)-(Name1-Rest1)
>0 with an inclusive mask of (Name1-Rest1)>0.
Given no objectives to stack the data in a single modell, what would be
wrong about simply calculating contrasts for post-pre data?
3. Even though the design might not have been designed to contrast
different response types of the aphasic patient (correct responses,
paraphasias & neologisms) their number in the 2 sessions (both taken
together) seems to be sufficient to contrast them...(which actually yields
neat results in frontal cortices). Are there any major theoretical
comments on a short TR (6 sec), which might leave us with some overlap
from previous items...
4. Does anyone has a good idea how to compare patient data with the few
control subjects beyond descriptive statistics...
thanx so much in advance
marcus
[log in to unmask]
|